Quiz (Chapter 10)
Inferences Concerning One or
Two Means
Introduction
 A ____ (small/large) number of statistical techniques are available
to researchers who want to deal inferentially with the mean(s) computed
from sample data.
 (T/F) The technique of interval estimation can be used in studies
where interest lies in one or two means.
Inferences Concerning a Single Mean
 When a single sample is taken from a population, at what will the
researcher be guessing, the sample mean or m?
 In general, large sample sizes tend to bring about _____ (narrow/wide)
confidence intervals.
 Look at Excerpt 10.4. Using symbols, indicate what the null hypothesis
was.
 With a single group, the null hypothesis can be rejected when the
difference between the sample mean and ____ is larger than what would
be expected by chance.
 the actual population mean
 the pinpoint number in the null hypothesis
 If a researcher conducts a ttest on a single group mean, and if
df = 19, how many subjects were in the sample?
Inferences Concerning Two Means
 Which of the following terms suggest that 2 samples are "correlated"?
(Choose 1 or more of the options)
 paired samples
 matched samples
 independent samples
 unpaired samples
 dependent samples
 If the two samples have different ns, this is a "tipoff" that the
samples are ____ (independent/correlated).
 If a group's pretest mean is compared against its posttest mean,
these two sets of data are considered to be _____ (independent/correlated).
 (T/F) In comparing 2 means with a statistical test, 2 separate
inference are made (one to m1,
the other to m2).
 Do these statements say the same thing?
 Ho: m1 = m2
 Ho: m1  m2
= 0
 (T/F) When comparing 2 means, Ho
does not have to be set up in a "no difference" fashion.
 In Excerpt 10.4, is the number 62 correct or is that number incorrect?
 The means of 2 independent samples can be contrasted via a ttest,
a __test, or a __test.
 What do the abbreviations df, SS, and MS stand for?
 In an ANOVA summary table, the calculated value (i.e., Fvalue) is
computed by dividing MSbetween by _____ .
 dfbetween
 dfwithin
 SSbetween
 SSwithin
 MSwithin
 In an ANOVA summary table, the total number of subjects in both groups
= dfbetween + dfwithin + ___ .
 How would you know that two sample means are significantly different
from each other if all you're able to look at are the two numerical
values that form the end points of the CI?
Multiple Dependent Variables
 If a researcher has data from 2 groups on 3 dependent variables,
and if he/she wants use a ttest to compare the means of the groups,
how many ttests would the researcher perform?
 1
 2
 3
 5
 6
 In Excerpt 10.15, the last number in the final sentence is .008.
What would this number be if there had been 2 rather than 3 disability
spheres?
 (T/F) The Bonferroni technique cannot be used when there's
only one sample.
 If the pseudoBonferroni technique is used by a ".05" researcher
who wants to compare two groups on each of 7 dependent variables, what
revised alpha level would he/she probably use?
 .01
 .007
 .001
Effect Size Assessment and Power Analyses
 When conducting a test on one or two means, do most researchers do
something to help them (and their readers) distinguish between statistical
and practical significance?
 In Excerpt 10.8, which number(s) should you look at if you're trying
to determine whether the difference between the two means has "practical
significance"?
 25.6 and 39.2
 103
 9.8
 .001
 Suppose a researcher determines that 20 subjects must be in each
sample in order for the statistical comparison of 2 sample means to
have a power of .80. In this study, there will be a 20% chance
of a Type __ error.
Underlying Assumptions
 What 4 underlying assumptions are associated with t, z, or Ftests
that compare 2 means?
 (T/F) If a researcher tests an underlying assumption, he/she hopes
to reject the assumption's null hypothesis.
 In Excerpt 10.14, the results of 5 ttests are reported. In which
of these ttests do the sample data support most the homogeneity of
variance assumption?
 What makes a statistical comparison of two means robust to the equal
variance assumption?
Comments
 In Excerpt 10.7, the statistical comparison of the two cohorts turned out to be
"p > .05." Can the null hypothesis of the ttest be accepted, since no
statistically significant difference was found?
 (T/F) The very small pvalue in 10.10 strongly suggests that all of the
teachers with prior learning about whole language literacy had higher scores than
did any of the teachers with no prior learning about whole language literacy.
 The last sentence of Chapter 10 says: "____________ by any of these
techniques, regardless of how bold the researchers' claims might be."
A Few Tougher Questions Over Chapter 10
 (T/F) Whereas a onetailed test must be used whenever
the null hypothesis deals with a single population, a twotailed
test must be used whenever the null hypothesis deals with two populations.
 In Excerpt 10.9, if there had been 50 women and 50 men involved
in the study, how many numbers in the excerpt would definitely change?
 In Excerpt 10.1, we are given the numerical values for the mean, the CI,
and the median. Which of those numbers is definitely not right?
 The mean is not right; it should have been larger than 60.
 The CI is not right; it should have been wider.
 The median is not right; it should have been somewhere between 54.7 and 58.5.
 None of the numbers in the Excerpt is definitely wrong.
 How should the concept of "overlapping distributions" influence how we interpret
the assertion in Excerpt 10.14 that "The hypertensive group had impaired performance
in all measures except Continuity of Attention"?
 Insert "slightly" after the word "had."
 Insert "significantly" after the word "had."
 Insert "clinically" after the word "had."
 Insert "On average," at the beginning of the sentence.
 If, in Excerpt 10.22, alpha had been set equal to .01 (rather than .05),
the power analysis would have indicated the need for ___ (more than/fewer than)
50 patients in each group.
 If a oneway analysis of variance is used to compare 2 groups of
20 subjects each, and if the SStotal turns out equal to 86
while the MSerror turns out equal to 2, how large would the
calculated value be?
 Look at Excerpt 10.23. In applying the KolmogorovSmirnov test, which of the
following alpha levels would have been most "conservative" in the sense that it
would have made it harder for the researchers to get a result that they wanted?
 .001
 .01
 .05
 .15
Click here for answers.
